Technology

Three solutions, one rule:data travels no further than it must.

At the counter, recognition runs inside the building. At home, it runs on the device. The eInk labels neither listen nor film: they receive a page that is already composed, and show it.

Block diagram

Where the work happens.

Three paths, three perimeters.

Counter station

Organisations: municipalities, healthcare, banks, post offices, pharmacies.

Perimeter: the building

  1. Station microphonesAudio captured at the desk, on both sides.
  2. Thin client8-inch display and 5-inch operator module.
  3. Building local networkNo route to the outside.
  4. On-premise server, NVIDIA CUDA platformRecognition, entity extraction, translation and summary.

All inference happens inside the building. No audio, video or text leaves it. No internet connection required.

Home device

Personal use, portable.

Perimeter: the device

  1. Integrated microphonesPicked up in the home.
  2. Edge AI, Qualcomm DragonWing platformModels run on board, on the neural processor: recognition and speaker segmentation.
  3. Screen and RGB LEDsText with one colour per voice, alerts for environmental sounds.

Audio never leaves the device. Integrated battery, no account, no network required.

eInk label

Care homes and shared rooms.

Perimeter: the service and the building

  1. The home's calendars and emailMenus, events, birthdays: the sources the staff already use.
  2. The service composes the pageIt picks the group's template and produces the image.
  3. The label shows itIt receives the image, draws it, and goes back to sleep.

The label computes nothing and has neither microphone nor camera: it shows what the home already pins to its noticeboard.

The AI part

Speech recognition, where it is needed.

Not every solution of ours carries a model: the eInk labels carry none. At the counter and at home they do, and the division of labour is always the same: the client captures and displays, it does not decide and does not store.

  1. Two clients, two implementations

    At the counter the client runs on ESP32, inside the two displays. At home it is a Linux application on the device itself. In both cases it does the same three things: capture audio and video (the video is what sign-language recognition needs), show the text, and accept a tap.

  2. Where there is a server, the models can be bigger

    The counter station has the on-premise server behind it, and that makes room for larger recognition models: fewer errors on the words, and punctuation in the right places. Punctuation is what makes a text readable instead of something to decipher.

  3. At home, everything on board

    On the home device the models are leaner and faster: the edge AI hardware runs recognition, speaker segmentation and, when needed, environmental sound recognition. Fewer parameters, lower latency, no server to reach.

  4. More than one model

    Recognition runs alongside speaker segmentation, entity extraction — addresses, phone numbers, amounts, dates — and translation. That is where the end-of-conversation summary comes from.

  5. Under evaluation: Auracast

    For museums, conference rooms and lecture halls we are evaluating Auracast (Bluetooth LE Audio): a single broadcast that anyone can pick up on a compatible hearing aid or on headphones — no pairing, and nobody has to ask for help.

  6. Models tuned to the context

    Models can be trained on the vocabulary that recurs in the place: forms, procedures and internal names are recognised better than a general-purpose model would manage.

Practical consequences

What a local architecture gets you.

They are about the counter and the home, where inference runs.

  1. No connectivity dependency

    The service does not stop when the line goes down, and it needs no guaranteed bandwidth on site.

  2. No third-party sharing

    There is no external provider receiving audio or text, so there is no contract to take on trust.

  3. Predictable latency

    The path from audio to text stays inside the local network, with no round trip to a remote service.

  4. Expandable ecosystem

    One on-premise server drives two to six stations: they are added as needs grow, and past six a second server joins in without rebuilding the installation.

The eInk labels

Electronic paper: pigment, not light.

The image is made of pigment reflecting the light of the room, like ink on a sheet, and it stays put with the power off. From there the division of labour is strict: the service prepares the page, the label shows it.

  1. The layout work happens upstream

    The page is composed once, on the server, and the label receives a finished image. Changing the design is a change on the server: there is no update to push to the devices.

  2. It draws power only when it changes

    Electronic paper takes current at the moment the image changes; between updates the device sleeps. That is what makes a battery enough, and lets the label hang where there is no socket.

  3. How they connect

    The labels use the home's Wi-Fi. For the places Wi-Fi does not reach — gardens, outbuildings — we are evaluating a LoRa Meshtastic network, which the devices build between themselves without leaning on the building's infrastructure.

  4. Ordinary parts

    The labels are built on ESP32 and Raspberry Pi Zero, with Spectra 6 colour screens: standard, documented, obtainable parts.

  5. The wall stays readable offline

    If the network drops, the label keeps showing the last page it received: electronic paper needs no power to stay visible.

The LoRa Meshtastic network is under evaluation: it is not part of what we install today.